Both types of romanization are meaningful in different contexts. For English pronouncing the words such as "Shinjuku", "Fukushima", "Chichibu" are better, and are better for writing Japanese names in English text in general. However, the other system, with "Sinzyuku", "Hukusima", etc is directly according to the kana by the grid order, so its use is if you need to directly correspond to such a thing.
Good point. And "Sinzyuku" and "Hukusima" not only correspond more closely to the kana; they also match most native Japanese speakers’ mental representations of the sounds. For example, most Japanese speakers perceive the し syllable as consisting of a single consonant phoneme followed by a single vowel phoneme, so si makes more intuitive sense to them than shi. And they perceive the consonants in は and ふ as being the same, so transcribing those kana as ha and hu seems more natural than ha and fu. (This doesn’t apply to zyu for じゅ, however; ju is better.)
When I was teaching English academic writing at a Japanese university fifteen years ago, I was struck at how inconsistently the students wrote Japanese names and terms in English. They had been exposed to multiple romanization methods, and most of them couldn’t stick to just one system. I ended up writing a set of detailed guidelines for them [1, 2].
An excellent source for even more information about the intricacies of writing Japanese in English is the Japan Style Sheet [3].
I have to disagree here. Hepburn is superior unless your goal is transliteration from Roman characters to kana. From the article (which also argues against kunrei)
> Romanized Japanese is essentially an auxiliary device — a reading aid, if you will — and not the real thing.
Romanization is for foreigners. Japanese people would read the original kana/kanji anyway
Korean went through a similar reform and I also disagree with it. Favoring transliteration (linguists and academics) over the actual sounds (everyday people) is harmful for communication.
> Romanization is for foreigners. Japanese people would read the original kana/kanji anyway
Assuming it is only used for reading purpose. It is actually also useful for typing in Japanese with a keyboard without kanas, though it is still mostly relevant for foreigners.
Side note: I personally like input systems that support both romanizations simultaneously, since there are shorter combinations in each one. For instance, "si" saves one letter compared to "shi", and "ju" is also easier to type compared to "zyu".
Korean Revised Romanization [1] also went forward with a similar logic two decades ago. It slightly prefers simpler rules for native speakers over the phonological accuracy, and it was a reasonable choice in my opinion.
The switch to Hepburn rules is reasonable since it's the de facto style of romanization but, if they were going to go thru this sort of effort to revise their written language it seems disappointing that they wouldn't add stress marks / accent marks to their romanization.
It especially makes sense in Japanese to remove ambiguity like, 港南. Konan, naively in Hepburn. But is it Kon-an or Ko-nan? Kònan removes ambiguity.
There's no ambiguity: in Hepburn, Konan is always "Ko-nan", you'd need to write it Kon'an if you wanted "Kon-an". Which is also unlikely in Japanese, unless you're writing about root vegetable safety 根案 or something.
If anything, using stress marks would be worse, because Japanese is (slightly) tonal and accents are on rare occasion used to mark them: háshi vs hashí.
It's こうなん, not こんあん, but apparently the hepburn romanization of こう (and other long vowels within the same morphene) is "supposed" to be represented by the accent macron (like "kō"), according to wikipedia:
Sounds like there's variation here, however -- see the picture of the railroad sign in the wikipedia article, which is a mishmash of different approaches.
I don't tend to dwell on romanization of words, because I just go to hiragana when there's doubt (which was my point).
Hepburn doesn't use the apostrophe for that. The apostrophe is for words with a glottal stop like ze'tai. Or without an apostrophe as zettai is possible too.
I used the Genki books which also use Hepburn. Simply put Hepburn mimics the English language and, the English language uses apostrophes for glottal stops. Especially if the language foreign. This is why Genki does that.
I think you have a very warped idea of how Hepburn works and your confusion is just caused by not knowing Hepburn. You would probably be better off actually looking up the rules of Hepburn instead of assuming that it's just the English language.
The Genki books seem to not use apostrophe but a dash, since I found a PDF of it and it uses the terms "on-yomi" and "kun-yomi" to distinguish んよ (n'yo / n-yo) and にょ (nyo). It also very clearly says that った becomes "tta" right in the beginning, so I have no idea where your apostrophe usage comes from.
港南 has a very unambiguous Hepburn romanisation of kōnan, which is impossible to confuse for こなん ("konan") or こんあん ("kon'an" / "kon-an").
zettai is not a glottal stop. Not sure what the term is, but it's a fricative of some sort. You're stopping the flow of air with your tongue on the back of your teeth.
A glottal stop is like "uh oh", where you stop the air with your glottis.
labiodental? err, i mean interdental? or is it alveolar? I know what you mean, like pausing half a step to put a barely perceptible extra bit of mass behind the release of the 't' sound.
Given that the "Ko" is actually pronounced "Kou" (i.e. hold the "o" sound for a beat longer), I'm not sure how the downward accent helps? The bigger thing than "kon an" vs. "ko nan", IMO, is that the romanization rules don't capture these kinds of repeated/extended vowel combinations.
Honestly, anyone who is sufficiently capable in japanese to know that 港南 is something close to "Ko + nan" is going to know hiragana and be able deduce the onyomi readings of each (or at least, of 南, which is the most common reading of a basic kanji), and will not need romanization. The real value of the romaji is for people who know nothing of the language, and it's just weird that the common romanization leaves these important vowel details out.
(I'd personally love a unified syntax for tone that doesn't involve drawing squiggly lines over the word, or a number next to it, but that's a pretty specialized need, I think.)
> I'd personally love a unified syntax for tone that doesn't involve drawing squiggly lines over the word, or a number next to it, but that's a pretty specialized need, I think.)
That would be lovely and possibly and attainable goal for a minimally tonal language like Japanese, but the challenge is generalizing any such system to heavily tonal languages like Hokkien, with six or seven tonal markings and additional layers of complexity like tonality rules that depend on different word boundaries and positions.
It's a shame that different shades of ink weren't invented and/or normalized sooner, because a lot of this state space could be captured using a simple convention of two stroke weights (light/heavy) by two shades of color (red black) for every glyph.
"Rubric" was an early gesture in this direction but unfortunately never escaped its original application.
Incorrect on both counts. 追う ou "to chase" is a simple example of separately pronounced consecutive vowels.
港南, on the other hand, does not have a "u" sound of any kind, it's simply a long "ō", which doesn't really exist in English. The reason it's occasionally spelled with a "u" (although not in any formal standard) is because hiragana uses the letter う to indicate long o and u.
My point was that you do not pronounce that word as "oh u". There's no glottal stop, nor are the vowels "separated". But I agree it's not a familiar concept to English speakers, nor is it easily expressed in romaji. If I were explaining it to a native English speaker who didn't know Japanese at all, I'd tell them to say "owe"...but that's not hepburn.
> 港南, on the other hand, does not have a "u" sound of any kind, it's simply a long "ō", which doesn't really exist in English.
Yes, that was exactly my point. It's a long vowel. The closest approximation in English is not "ko", but "kou".
I disagree that this isn't in English, however. It's an easy concept to map -- we have tons of repeated vowels -- which is why "kounan" gets the pronunciation pretty close.
Directly addressing the question I was responding to, there's no question about "ko un an" vs. "kou nan", because the former requires you to stop the airflow between the "ko" and "un" in a way that isn't natural for Japanese. Knowing that consecutive vowels will essentially always "smoosh together" allows you to pronounce the word -- so long as you know the vowel is long.
I think this makes sense given the prominence and place English holds in the world.
Hepburn has made Japanese more accessible universally given its phonetic nature and relationship to English. There are many other transliteration systems out there that people are using with their local languages as well.
If you want to see the opposite approach, look at China and Pinyin. It’s made pronunciation of Chinese in English and elsewhere less intuitive. In order to pronounce Chinese words properly you need to learn Pinyin and its phonetic rules. As a result, many Chinese people take on Westernized names to avoid the inevitable butchering, which is a bit of a shame.
While Pinyin was probably the best choice for enhancing literacy in China, it has inadvertently created hurdles for foreigners. In contrast, Japan’s Hepburn system has avoided such challenges.
> If you want to see the opposite approach, look at China and Pinyin. It’s made pronunciation of Chinese in English and elsewhere less intuitive. In order to pronounce Chinese words properly you need to learn Pinyin and its phonetic rules. As a result, many Chinese people take on Westernized names to avoid the inevitable butchering, which is a bit of a shame.
Taiwanese romanization is way better. It's much closer to English and more readable
Chinese people take Western names because Chinese are very flexible and, especially, because the Chinese naming system is quite alien to Westerners.
There is also the issue of tones, which is the major hurdle for Westerners.
Now as to "you need to learn pinyin and it's rules". Yes, you need to learn and make some effort to pronounce a foreign language. The same goes for any language and it is unavoidable.
sure, maybe for english speakers an english based transliteration might be better. but i'd argue that makes it worse for everyone else.
given how irregular english pronunciation is, i find that a transliteration system is better if it is based on languages with more regular pronunciation of letters.
i don't see how chinese pinyin creates any more hurdles for foreigners in china than say german does for foreigners in germany. (to be fair, i am fluent in both english and german, so maybe i am not a good judge for this)
> If you disagree, show me one language whose speakers would naturally pronounce "Sinzyuku" correctly.
Not one language has speakers who will naturally pronounce "Shinjuku" exactly correctly either - no two languages are alike, and it's normal and healthy to have pronunciation differences between languages natively written with the same alphabet (e.g. English speakers will naturally mispronounce many Spanish words - no-one would suggest this means Spanish spelling should change). Plenty of languages (e.g. Irish or Polish) naturally read "Sinzyuku" as something closer to the correct pronunciation than "Shinjuku", which is the only reasonable way to compare.
There's hiragana/katakana for regular transcription of spoken Japanese. I'm guessing that's slightly different situation to Chinese language which does not have a phonographic script.
That sounds backwards to me. Hepburn is widespread because Japan once had a very widespread reach over the world, so much that a system tuned for westerners is inevitably going to appear. Chinese Pinyin is a much later development and widely known Chinese names are indeed written in Pinyin with minor variantions. For example, the name "Beijing" is an unaccented version of Pinyin "Běijīng", and it swiftly replaced a previous spelling "Peking" which was in use for over 4 centuries.
> Hepburn has made Japanese more accessible universally given its phonetic nature and relationship to English.
It's made Japanese more accessible to English-speakers, at the cost of making it less accessible to those from other languages, and making romanised Japanese harder for Japanese people to write, and perhaps even indirectly impeding Japanese learning of English (which is notoriously terrible) as well. This is perhaps a good tradeoff for tourist-oriented materials, but I'm sad to see it becoming the standard.
> As a result, many Chinese people take on Westernized names to avoid the inevitable butchering, which is a bit of a shame.
Plenty of Japanese names are also butchered, whichever romanisation system you use. E.g. "Abe" is romanized that way in either system; "Mazda" and "Ohtani" are deliberately romanized "wrong" (again under either system) for English speakers to mangle them less. IMO taking up a name that's natural in the culture you're operating ends up better for everyone. (I very happily use a Japanese name in Japan rather than trying to get Japanese people to pronounce my birth name).
How is it harder for non-English speakers? Knowing the kana pronunciation for Sinzyuku is way harder than knowing the English pronunciation of Shinjuku for most outside of JP. Also, JP schools already teach English from grade 1 onward so if anything, seeing the English transliteration aligns way better and enforces what they are already being taught. Not to mention katakana spellings and pronunciation which further confuse their learning of English. Previously you effectively had three different representations of foreign words with different spellings, this cuts it down to two which significantly decreases the cognitive overload
> How is it harder for non-English speakers? Knowing the kana pronunciation for Sinzyuku is way harder than knowing the English pronunciation of Shinjuku for most outside of JP.
Only if they're English speakers! If your only language is e.g. Irish then writing that sound as "Shi" is more confusing than writing it as "Si".
> Not to mention katakana spellings and pronunciation which further confuse their learning of English. Previously you effectively had three different representations of foreign words with different spellings, this cuts it down to two which significantly decreases the cognitive overload
Maybe. Or maybe the way Hepburn creates "blurred" words that are partially English and partially Japanese causes more confusion than having a clear distinction between written English and written Japanese words that makes it easier to pronounce each one in the way appropriate to that language (i.e. fully code-switch as fluent bilingual people do).
Yeah those are fair points. I still think that to encourage wider learning and understanding from foreigners, using English as a base model makes sense because it is most widely taught across the world.
As far as the second point, I taught English in Japan and found that while English education starts at an extremely young age, their relative skill is pretty low compared to other countries that do the same. There are a couple of reasons for this but I think further integration of English pronunciation could help in this department while also appealing to the widest swathe of tourists possible.
Another funny thing I’ve found is that the romaji JP iphone keyboard autofills kana and kanji best with hepburn inputs
Kun-rei has a learning cost and no one is paying for it. Hepburn or its close-enough "Ohtani" analogue is free due to (notoriously bad)compulsory English training as well as use of Romaji input on computers, and technical rule conformance to actual Hepburn is never questioned so it wins.
I don't know, I think the real sad part is that everyone knows the casual Hepburn from Romaji popularity; Kana shall be default from a nationalistic, purist, standpoint. Mazda and Ohtani are fine, certainly better than Zhang and Nguyen situation, which are supposed to be pronounced "chan" and "gwen" but being mostly sources of confusion.
I mean it probably helps that few, if any Japanese sounds are alien to English speakers. I've studied some Japanese, been several times, can't remember any sounds that were particularly difficult (whereas Japanese speakers definitely struggle with some English sounds, like "th" and "l").
There are intonation aspects that are hard for (native) English speakers, and the syllable speed is higher than English, but the basic sounds are straightforward to grasp.
つ is an unnatural sound for at least some English speakers, and can be hard to distinguish from す. The short/long vowel distinction is also completely alien at first (e.g. people are mostly familiar with the pronunciation of Tōkyō, but the natural reading of "Tokyo" in English would be quite different, starting with a "tock" sound).
Interesting writeup(s) about the popular culture reasons and the more concrete technical reasons for the change.
Curious though of this move to more English-favoring mode when it seems in recent years in other countries as far as naming etc there was a move away from English/colonial/etc naming of places etc like Kiev to Kyiv (sorry that one's from Russian), and like places in India awhile back.... Similarly more use of Indigenous naming etc just off top of my head. I suppose that's formal naming and not language constructs but still.
This is basically a case of the government removing a rule that almost nobody, including them, was really following. I guess they had it written down somewhere to use Kunrei, but like TFA says Hepburn is what's used virtually everywhere.
So, whatever system one might expect ought to be truer to the source language, the Anglicized system is the one that current native speakers find most familiar.
I recognize Kunrei-shiki as the style used in Eleanor Harz Jorden's seminal Japanese textbook, Japanese: The Spoken Language. I always thought Hepburn style was more appropriate for the goals of that book: to convey spoken Japanese in a way understandable to English speakers. But I think Jorden intended for accompanying audio materials to supply a pronunciation guide. (My Japanese professor was a native speaker, and so used her own voice, not audio materials accompanying the textbook.)
Good. In practice the old system was taught in schools, but not really used officially. Except for people giving their names, where someones name is definitely pronounced as Shota, they would write it Syota. I just checked facebook and found quite a few people like this.
The "n" in "na" is not the same as the "n" in "ni".
However, only some consonants get this treatment, such as "sa" and "shi", because it only takes into account what a typical English speaker can distinguish.
Do they have to keep such an English-centric approach? Why not "nyi"?
The article says Hepburn romanization is based on English consonants readings. Maybe if they based it on Russian instead you would see them using iotated and non-iotated vowels.
But it uses Roman letters which form the Latin alphabet.
Romanization here means that Japanese text gets transcribed into Roman characters. The fact that the transcription follows English phonetics is a different thing.
I wish Korea would do the same. I've taught in linguistics departments in Japan and Korea, and I've had many arguments over romanization policies.
What is the point of romanization? Who is it for? If it is for the benefit of non-speakers, then the pronunciation assigned to the Latin letters should be predictable based on common usage internationally. If it is for the benefit of native speakers...well, why? Japanese and Korean have perfectly good alphabets. It doesn't benefit Japanese or Koreans to transcribe their language into the Latin alphabet.
Pinyin is a different case. As I understand it, Pinyin was developed in the 1950s as a literacy aid for native Chinese speakers, so developers were free to be creative with the pronunciations of the letters.
Sign of the worldwide dominance of English, sadly. Hepburn is utter nonsense if your native language is e.g. German or Irish - but it's popular with English monoglots who believe that their peculiar use of the Latin alphabet is somehow objectively correct, and they seem to be in ascendence these days.
Hepburn is not equal to the English spelling rules though, which are rife with inconsistencies and exceptions. Even though it doesn't match with how many language's version of the Latin alphabet works, and one might object to that, Hepburn an effective tool to get the job done. English learners wish that the rules of English orthography were that consistent!
If one's goal is to travel to Japan or to research their culture, knowledge of Japanese and/or English is required anyways. In the latter case, one will be positively surprised that there are only a few rules specific to Japanese to be learned to be able to read with some degree of accuracy.
One might object to the dominance of English as a lingua franca, but actually any other languages being the lingua franca would dominate the Romanization system as well, and that would lead to similar issues for speakers whose writing system works differently.
> One might object to the dominance of English as a lingua franca, but actually any other languages being the lingua franca would dominate the Romanization system as well
Kunreisiki is not a system dominated by any particular lingua franca; it's a system that directly reflects Japanese ways of writing Japanese. One could imagine e.g. a Spanish-influenced romanization system that would represent the は-line in a way that corresponded to Spanish orthography becoming popular in a world that was dominated by Spanish as a lingua franca, and that would have the same problems that Hepburn does. But Kunreisiki really is "neutral" in that sense.
In short, it seems to make the Latin alphabet work for Japanese. Perfectly fine if Japan uses it as its Latin-based writing system (discarding the native ones), or to transcribe it for academic usage, but unhelpful as a Romanization, which is mostly intended to make it easier for people completely unfamiliar with the language to pronounce names and to learn the language.
Hepburn may not be an ideal match for German, but it's still a better one than Kunrei-shiki. I'm fairly sure there is no language in the world whose pronunciation of Latin letters matches the latter (hah!).
The only thing Kunrei-shiki has going for it is that it's more regular and can be more easily matched to kana. It's easier for Japanese kids to learn. But they are not the ones who are going to use it primarily.
You could use different romanization systems for each foreign language, and there are some (there is one for German where あいち is "Aitschi"), but that doesn't end up serving anyone well in a globalized world with an Internet.
The Latin alphabet does not include J, U, and W, but the English alphabet does (see "Shinjuku") and this would likely be better understood as a form of Japanese spelling reform rather than "Romanization".
> The Latin alphabet does not include J, U, and W, but the English alphabet does (see "Shinjuku")
Japanese does not naturally include distinct U/V letters and does not have a first-class concept that needs to be written with a J; the official way to romanize that place name is "Sinzyuku" and that's morally equivalent to a traditional-Latin "SINZYVKV". W does exist but is decidedly second-class and declining (only the "Wa" sound is really heard any more; I suspect it will go the way of obsolete phonemes like "Ye" and "We" soon enough).
> this would likely be better understood as a form of Japanese spelling reform rather than "Romanization".
It makes no sense as a way of spelling Japanese. It's about standardising on anglicising Japanese instead of romanising it.
My native language is Irish and I disagree. If you think English pronunciation rules are odd or inconsistent, then Irish will make you cry out in pain, as dialects from less than a hundred miles away can sound wildly different.
I don't use Romaji much but I learned the Japanese syllabary ~35 years ago so translating Hepburn to correct pronunciation doesn't involve any conscious thought on my part.
> My native language is Irish and I disagree. If you think English pronunciation rules are odd or inconsistent, then Irish will make you cry out in pain, as dialects from less than a hundred miles away can sound wildly different.
Irish pronunciation may be local, but the orthography is very consistent. I actually felt there was a lot in common between Kunreisiki and Irish - at first blush it's very confusing for a native English speaker, but once you get over that initial reflex and read it according to its own rules it all makes sense.
> I don't use Romaji much but I learned the Japanese syllabary ~35 years ago so translating Hepburn to correct pronunciation doesn't involve any conscious thought on my part.
Right, but doing that will paper over the issues with any romanisation system. What do you find worse about Kunreisiki?
I don't have a strong opinion on Kunreisiki; when I first came across it I thought 'this looks needlessly complicated' and just assumed it was old and outmoded so basically I've always ignored it. My first introduction to Japanese was from a friend who taught me the pronunciation and hiragana at the same time, so I got used to the sounds well before encountering Romaji. I only learned about the history of Romanization and took a more rigorous interest in the language years later.
That means terms like “shinjuku” will probably survive as retro touchstones - good vaporwave material, when people look back and say “remember when sinzuku used be spelled ‘shinjuku’ and it sounded like…”
You've got it backwards. Shinjuku is sticking around as the official spelling, or rather becoming the official spelling by adopting Hepburn romanization. The pronunciation wouldn't be any different in either case.
It’s not parent. This article is either terribly written or horribly translated. The captions in the photos don’t match what said in the text and aren’t logically consistent across the article.
Picture of Shibuya Station's glowing sign. Caption: "the famous Tokyo shopping district known worldwide as Shibuya will be changed in its official presentation from Sibuya"
Currently, Shibuya is officially supposed to be romanized as Sibuya, but in common usage including signage it is written Shibuya.
"Amid concern the divide between official rules and common usage is causing confusion, a subcommittee of the Council for Cultural Affairs deemed it necessary to consider the revision to improve communication."
Since they want to adapt the law to match what everyone is already doing anyways, no change is to be expected.
It may technically be logically consistent, but nobody writes this way: “meaning, for example, the official spelling of the central Japan prefecture of Aichi will replace Aiti.”
This reads that the official spelling is currently Aichi.
The article talks about how "y" is replacing "x" as the official romanization in Japan. "The official spelling of the central Japan prefecture of Aichi" indicates that "Aichi" is the official spelling. This means to me that "Aichi" is "y" and, by extension, that "Aiti" is "x".
"Aichi will replace Aiti" doesn't add any clarity if it's breaking that logic. There's a reason why others also find this article hard to parse.
It's hard to parse because the Japanese system is itself a mess. The (former) official romanization rules say the name of this prefecture should be written Aiti but the official practice is to write it as Aichi because that's easier to teach to non-Japanese.
Consider that the romanization rules were (iirc) mostly originating from the ministry of education, but that ministries responsible for trade and tourism would naturally prefer to use whatever is easiest for foreign partners to deal with, and could lobby for exceptions since they're revenue centers rather than cost centers. Japan has both a professional bureaucracy under a parliamentary system and de facto one party rule, so (imho) there's much greater potential for policy balkanization of this sort.
People absolutely do write this way. My guess is that the article writer is from the UK rather than the US. British English often involves longer and more complex clauses and something not unlike operator overloading.
It's logically consistent, but clunky to read and prone to misunderstanding. I had to read over it a couple of times to understand, are they changing from Sibuya to Shibuya or the other way around?
As an example of text written by native English speakers that has these properties of being entirely logically consistent but clunky and difficult to understand, see the Second Amendment to the U.S. Constitution.
Not knocking him, just clarifying. It is terribly written (The Japan Times is an English-language Japanese publication, you can safely knock them for it), but the (untranslated) chart tells the tale: Kunrei-shiki on the left and Hepburn-shiki on the right. What the article doesn't make clear is that Hepburn-shiki is what we're already used to seeing overseas, so for us, nothing is changing. Shibuya is still Shibuya, but the written Romanization is not the best pronunciation guide no matter what form of Romaji.
I completely misread the article too (and the 'Additional context' article posted here), I think because it words things strangely but also the whole message is about change and all the examples seem to be of places where the well known spelling is already the Hepburn spelling. So the change is to official spellings that aren't used?
>So the change is to official spellings that aren't used?
Yes, the whole point is that kunreishiki is not practically used so it doesn't make sense for it to continue to be the official way to romanize things and for it to continue to be taught in school when no one is using it.
Kind of. The Japan Times is interesting. It's an English-language publication, but it is written for Japanese people by Japanese people. Despite their target audience being Japanese people, the fact that it is an English-language publication kind of makes it a default paper (well, figurative paper) of record for what's going on in Japan for people outside of Japan that don't really speak Japanese.
They also make textbooks for Japanese-language learners.
I don't know about that. If you look at the bylines more than half the writers come from anglophone countries, and a lot of the feature articles seem aimed at foreign settlers in Japan. They take breaking news from the Kyodo newswire and mostly do features.
My Japanese isn't good enough to read the news at speed, but I can manage TV news; there's a lot of stories that don't show up in the Japan Times until a week later. They do cover major political, economic, and crime (if it's a murder) news in a timely fashion.
It is possible my information about The Japan Times is basically historic and out of date at this point since organizations do change. This is the premise upon which the paper was originally founded and continued to operate on for most of its history, but I also don’t follow The Japan Times anymore.
Came here to say exactly this, the article is written maddeningly.
English readers will more naturally expect the "before" to come first in a sentence, and then the "after", but in the article the "before" consistently comes at the end of sentences.
The government officially used the Kunrei-shiki romanization rules for decades (apparently, though this is the first time I've ever heard of it), but is soon changing to using the Hepburn romanization rules [which correspond to nearly 100% of romanized Japanese terms that anybody would ever have seen in the wild].
The "diagram" in the article doesn't help English-speaking audiences at all, so it's out of place in this article. It seems to have been lifted directly from some other source, and what's more it's kind of a badly-designed diagram to begin with: the header reads something like "change in romanization system", each of the green background terms are names, and under it on the left is that name romanized under the old system and on the right is under the new system... but the names of the systems are in the first data row instead of in the header row / a second header row. Just.. so confusing.
The article is a bit confusing, and the photo with the station's name doesn't help,
The station names and place-name signs, when we see them written with Latin characters, are using romanization Hepburn rules since many decades ago, shinjuku stay shinjuku.
What is happening is that at the same time, the Ministry of Education had been officially teaching different romanisation rules to the Japanese, denominated Monbusho system, which is slightly different with some latin syllables, like si (shi in Herpburn) or ti (chi in Herpburn), but now they are officially going to step towards Herpburn rules, the same as those used in the place-name signs when are written with Latin characters.
